Booted Windows with cpu fan off - how bad is this???

[marlinman]

I'm pretty sure I need to replace my Athlon 64 3200+.

My PC restarts itself all the time under xp64 (although suspiciously NOT when I play SupCom for hours on end...). On a different hdd with vista32 installed I can run Prime95 for hours on end with no probs... while at other times errors are generated within seconds. I've yet to correlate this behaviour with the particular test being run.

I've run memtest std for 8 hours straight with no errors so think the Crucial ballistix are ok. (Plus I've never given them more than their spec'd 2.8v).

I'm using a dfi nf4 sli infinity m/b. The only stress I've subjected this to is raising the chipset voltage briefly (and not excessively) to see what effect this had on benchmarks.

Interestingly, my DVD writer fails to accurately burn stuff at the moment tho' when attached to a different machine it goes fine.

Oh yeah - and as the title says I stupidly booted xp64 with the cpu fan jammed by a cable and suspect this won't have done much good. I'm using a Zalman CNPS7000B-ALCU cooler so it could have been worse...

Plus: back in the day when I could overclock this chip without probs I ran it pretty close to the 50C ceiling - and occasionally a little over this - but for a matter of weeks and not months...

AND (!) my heatsink has a nice scratch across it thanks to another brilliant move on my part. Perhaps this leads to hotspots (or a hot 'line')...

While I'm here: any comments re the overclockability of the san diego core a64 3700+ appreciated! Otherwise I'll just get another venice 3200+ or maybe 3500+...

 

[Mallet]

For the DVD writer... Have you tried updating your ATA/IDE drivers?

 

[lookin4dlz]

Do they have thermal throttling like Intel chips? I've briefly turned on my computer a couple of times w/ Intel inside & no heatsink/fan and no problems w/ the chip over 2-3 years of 24/7overclocking use. I assume throttling kept it from mishap, so maybe your chip was protected as well.

 

[GuitarDaddy]

I highly doubt your CPU is damaged. I would
1. reset the CMOS
2. repair install of windows
3. reinstall any device drivers that are causing problems

 

[marlinman]

well I've reset the cmos and things seem to have got worse

I keep getting bluescreens when attempting to boot off the vista and xp install cds (GuitarDaddy: thought I'd just reinstall rather than repair)

PAGE_FAULT_IN_NONPAGED_AREA is the message that commonly appears

I gather this implies some kind of hardware failure ... great!


edit: solved! I tried upping my ram voltage to 2.8v and all is good... which is weird as before I got into overclocking it was running fine at the voltage it refused to work at a few hours ago... now to install an os and try prime95 again...

 

[marlinman]

ok I've reset the cmos, fiddled with date/time/vdimm, and reinstalled vista. No audio driver yet for my dfi nf4 sli infinity (strangely this isn't in the nf4 driver package for vista but is in the xp/xp64 nf4 package) but this is I think a problem for another time!

Now I'm STILL unable to get prime95 to run ok - it either brings vista down completely or just returns an error message.

What problems is this symptomatic of? I'm guessing cpu/mb, and as I said above I've done little to the mb to foul it up.

While I'm here - am I following a recommended (software) route to diagnosing a bad cpu or is there a better way? I have no other compatible cpu/mb to play with sadly.

 

[SimMike2]

I think AMD64 on up have good heat protection/shutdown options. A couple of years ago, one side of my heatsink clamps failed, thus causing the heatsink to lift totally off the CPU. I noticed this when I came home and computer was off. On starting, it got all the way to XP and was running OK. I had speedfan running and immediately saw the temperature sky-rocketing. No doubt it would have shut down again, but I shut it down from the start menu and discovered the problem. After fixing this, this CPU has performed flawlessly for at least two years.

 

[3NF]

What CPU voltage are you using? Increase it.

 

[marlinman]

3nf: your suggestion filled me with hope!

sadly it has not solved my problem

I've upped vcore from 1.4v to 1.525v but prime95 still barfs within seconds. This is under vista32 and xp64 (both fresh installs).

Interestingly, cpuz under vista reports a core speed of 1GHz which occasionally spikes to 2GHz (!). As well as a vcore of 1.1v, mult of 5x, and other rubbish. Under xp64 it reports accurately tho'. (I disabled cpu spectrum spread in bios as well as upping vcore)

I'm really at a loss with this one...

Just how does one go about diagnosing a screwed cpu?

 

[betasub]

Originally posted by: marlinman
Interestingly, cpuz under vista reports a core speed of 1GHz which occasionally spikes to 2GHz (!). As well as a vcore of 1.1v, mult of 5x, and other rubbish.

Cool&Quiet: power saving feature that reduces Vcore to 1.1V, and multiplier to 5x, during periods of CPU inactivity.

 

[3NF]

Originally posted by: betasub

Originally posted by: marlinman
Interestingly, cpuz under vista reports a core speed of 1GHz which occasionally spikes to 2GHz (!). As well as a vcore of 1.1v, mult of 5x, and other rubbish.

Cool&Quiet: power saving feature that reduces Vcore to 1.1V, and multiplier to 5x, during periods of CPU inactivity.

I would turn the Cool & Quiet off

 

[RMSe17]

Originally posted by: marlinman


PAGE_FAULT_IN_NONPAGED_AREA is the message that commonly appears

I gather this implies some kind of hardware failure ... great!

That's RAM related. I got that when I fried my RAM. Interestingly, the RAM gave 100% on memtest86+, but would error on Prime95 instantly. When I removed 1 stick, it ran for 4 hours fine, until I stopped it, when I switched to the other stick, it dumped out immediately. So one of my sticks was bad. (then my 2nd went bad too... after couple weeks of usage with just 1 stick)

 

[marlinman]

Thanks for all the ideas folks - this hard nut will be cracked dammit!

CnQ off but still crashing...

RMSe17: altho' I cleared up the problem you refer to by upping vdimm i'm trying just one stick now and prime95 has been running 10 minutes woohoo!

So IF the ram is to blame ( I use 2x512 crucial ballistix ddr400 pc3200 2-2-2-6 ) and given that they've never got more than their spec 2.8v (although they have been subjected of course to >ddr400) and given also that they're 8 months old and are "overclocker's ram" how much luck am I likely to have getting my supplier to replace?

 

[nyker96]

I thought modern CPUs got heat protection circuits and shuts down before damage. Maybe just didn't do it soon enough in this case.

 

[myocardia]

Originally posted by: marlinman
So IF the ram is to blame ( I use 2x512 crucial ballistix ddr400 pc3200 2-2-2-6 ) and given that they've never got more than their spec 2.8v (although they have been subjected of course to >ddr400) and given also that they're 8 months old and are "overclocker's ram" how much luck am I likely to have getting my supplier to replace?

The RAM isn't to blame, you just don't know how to setup your motherboard to work with it. Set your vdimm to 2.80v, set your tREF to 1560, set your tRC to Auto, and your tRFC to Auto, your DRAM Drive Strength to Level 8, and your DRAM Data Drive Strength to Level 4 (Reduce 00%). If you aren't overclocking, you can leave the DRAM Response Time on Fast; if you are overclocking the RAM, put it on Normal. Oh, and suppliers aren't who you send bad products back to. You send back products back to the manufacturer.

 

[marlinman]

aaaah... myocardia... I've missed your uniquely abrasive contributions...

The RAM does indeed seem to be to blame: following RMSe17's advice I've tested both sticks and found one wanting. First, I reset the cmos yet again, disabled cool'n'quiet, and adjusted date/time. No other changes made. I know the RAM worked at 2.6v when I got it... and felt uncomfortable when as detailed above I had to up vdimm to 2.8v to get OS's to boot off install discs! So vdimm is now 2.6v again. The 'bad' stick fails to boot xp64 or vista32 but booted xp32 (no service pack). Prime95 fails pretty much immediately tho' under xp32. With the 'good' stick (in the same slot as the 'bad' used) all is fine - at least I'm waiting for p95 to die (if indeed it does).

Thank you RMSe17 - you rock and I offer to have your babies!!! Thanks also to everyone else who contributed - this board has once again bettered me no-end!

No thanks to memtest's inability to demonstrate the faultiness of my dimm!

 

[myocardia]

Originally posted by: marlinman
aaaah... myocardia... I've missed your uniquely abrasive contributions...

Well, like I said to start with, you just don't understand how to operate that RAM with the motherboard you have. The fact that you don't like an answer, doesn't make it wrong. Crucial won't even consider taking it back, unless it refuses to run @ 2-2-2-6 timings, with 2.80v of vdimm. But, you're welcome to buy new RAM, if giving yours the amount of vdimm that's specified by the manufacturer (for it to run @ 200 Mhz) skeers you.